Geometrical descriptors for human face morphological analysis and recognition

详细信息	查看全文 | 推荐本文 |

• 作者：Enrico Vezzetti ; ^{enrico.vezzetti@polito.it}  ; [Author Vitae] ; Federica Marcolin  ; [Author Vitae]
• 关键词：Geometrical human modelling ; Face morphometry ; 3D scanners ; Geometrical descriptors ; Differential geometry
• 刊名：Robotics and Autonomous Systems
• 出版年：2012
• 期刊代码：153_09218890
• 类别：et
• 出版时间：June, 2012
• 卷：60
• 期：6
• 页码：928-939
• 文件大小：4329 K

摘要

The face is one of the most important parts of the human anatomy, and its study is very important, especially for developing automatic public security recognition strategies. In order to support this field, it is necessary to find a formal way of converting what the human eyes normally do in recognizing one person from another by extracting implicitly some morphological features. Since human recognition happens through an automatic 鈥渁uthentication鈥?of facial shape and features, this study should be undertaken in the geometrical domain.

The technical literature shows many parameters that could be adopted for finding a solution to this problem, but at present there is no evidence of a reliable solution. For this reason, this work, analysing strengths and constraints of what is available in the geometrical domain, provides the first guideline for supporting the development of an automatic face recognition approach.

Starting from differential geometry, such as the coefficients of the fundamental forms, the principal curvatures, mean and Gaussian curvatures, the derivatives and the shape and curvedness indices introduced by Koenderink and VanDoorn, this paper, working on a consistent set of case studies, analyses the geometrical descriptors鈥?performances in the three-dimensional facial study by the use of a set of indicators (similarity between different faces, sensitivity to noise, etc.). This is a preliminary study for analysing the behaviours of these descriptors on faces. It may be used as a guideline or a theoretical framework for researchers studying face shape or for face recognition applications.